Using Drones to Recognize Human Actions with Neural Networks
Author Information
Author(s): Abbas Yawar, Al Mudawi Naif, Alabdullah Bayan, Sadiq Touseef, Algarni Asaad, Rahman Hameedur, Jalal Ahmad
Primary Institution: Air University, Islamabad, Pakistan
Hypothesis
Can a neural network model effectively recognize human actions from drone-captured RGB videos?
Conclusion
The proposed system achieved action recognition accuracies of 0.68, 0.75, and 0.83 on three benchmark datasets.
Supporting Evidence
- The proposed method shows improved action recognition accuracy compared to previous techniques.
- Key skeletal points were extracted to enhance action classification.
- YOLOv9 was used for effective human detection in drone footage.
- Kernel Discriminant Analysis was employed to optimize feature extraction.
- The system was validated on three benchmark datasets: UAV-Human, UCF, and Drone-Action.
Takeaway
This study shows how drones can help recognize what people are doing by using special computer programs that look at videos.
Methodology
The study involved segmenting drone-captured videos into frames, preprocessing them, detecting humans using YOLO, extracting key points, and classifying actions with a CNN.
Limitations
The study may be limited by the quality of the datasets and the complexity of real-world environments.
Participant Demographics
119 individuals participated in the UAV-Human dataset.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website